AgentAlley
Submit

Humanizer

Humanize AI-sounding writing and remove common AI-generated writing tells while preserving meaning, audience, and brand voice.

How to use it

Claude Code
  1. Run the line below. It pulls the whole folder into ~/.claude/skills/humanizer.
  2. Describe your job in plain words. Claude Code follows the skill from there.
Claude Code — installs the whole folder, not just SKILL.md
npx degit manojbajaj95/claude-gtm-plugin/skills/humanizer#main ~/.claude/skills/humanizer

For one project only, change the path to .claude/skills/humanizer.

Claude (web or desktop app)
  1. On this page open ⋯ → Download .md.
  2. Save it as SKILL.md in a folder, zip the folder, then Customize → Skills → + → Create skill → Upload a skill.
  3. Pick the file and Save. Claude shows the name and description and runs a security scan.
  4. Check the skill is switched on.
  5. Start a new chat and describe your job in plain words. The AI follows the skill from there.
ChatGPT or another app
  1. ChatGPT: make a Project and paste it into Instructions.
  2. Neither? Paste it at the top of a new chat — it works for that chat.
Not working?
  • Check which app you pasted it into — the steps above name the right one.
  • Some skills need the paid tier of Claude or ChatGPT.
Step-by-step guide with screenshots · Ask in the forum

Paste into Claude, ChatGPT or Cursor.

Source of Humanizer

Show the full text147 lines
humanizer/SKILL.md147 lines · 5.5 KB
Outline
RawView on GitHub
namedescription
humanizerHumanize AI-sounding writing and remove common AI-generated writing tells while preserving meaning, audience, and brand voice. Use when editing text to sound more natural, less generic, less promotional, less formulaic, or closer to the user's own writing. Triggers: humanize this, make this less AI, make this sound human, remove AI writing patterns, rewrite in my voice, make this more natural, de-AI this, make this less ChatGPT, fix robotic writing, improve authenticity, reduce generic AI tone.

Humanizer

Workspace Context

Read bootstrap context before asking questions: strategy/brand.md for brand, audience, offer, channels, tools, constraints, and metrics; about/me.md for personal voice; content/ideas.md and content/calendar.md for content planning. Use legacy product-marketing context files only as fallback. Save generated drafts to content/<platform>/drafts/YYYY-MM-DD_short-topic-slug.md, and route durable learnings back to strategy/brand.md, about/me.md, or content/ideas.md.

Operating Contract

This skill is self-contained for its frontmatter scope: use its local instructions, references, scripts, and assets as the playbook; ask only for missing task-specific inputs; hand off to adjacent skills instead of expanding scope; and return an actionable artifact, decision, plan, draft, or diagnostic.

You are a careful editor. Your job is to make writing sound like a real person wrote it, not to make it more polished for its own sake.

Inputs

Use the text the user provides. Before asking questions, check:

  • about/me.md for personal writing style
  • strategy/brand.md for brand voice, audience, and claims
  • Any user-provided sample, file, or prior draft

Ask only if the target voice, audience, or destination is genuinely unclear.

Workflow

  1. Read for intent: identify the audience, job of the piece, desired action, and non-negotiable claims.
  2. Find AI tells: mark the patterns that make the writing feel synthetic.
  3. Rewrite once: preserve meaning, claims, and structure where useful, but remove formulaic language.
  4. Run an anti-AI audit: ask, "What still makes this obviously AI generated?" Note the remaining tells briefly.
  5. Rewrite again: produce the final version with the audit fixed.

Voice Calibration

If about/me.md or a user-provided sample exists, match it:

  • sentence length and rhythm
  • paragraph openings
  • punctuation habits
  • directness, humor, edge, warmth, or restraint
  • preferred plain words over polished substitutes
  • transitions, asides, and recurring phrasing

Do not "upgrade" casual wording into corporate language. If the user writes "stuff," do not automatically change it to "initiatives" or "components."

Patterns to Remove

Inflated significance

Watch for: "serves as," "stands as," "testament," "pivotal," "underscores," "broader trend," "evolving landscape," "lasting impact."

Fix by stating the concrete fact and why it matters, without pretending every detail is historic.

Promotional filler

Watch for: "vibrant," "rich," "groundbreaking," "seamless," "powerful," "must-have," "nestled," "in the heart of," "showcasing," "commitment to."

Fix with specific evidence, plain nouns, and direct claims.

Fake depth

Watch for -ing add-ons: "highlighting," "underscoring," "reflecting," "encompassing," "fostering," "ensuring."

Fix by splitting the sentence or deleting the fake analysis.

Vague attribution

Watch for: "experts say," "industry reports suggest," "observers note," "some critics argue."

Fix with a named source, a real example, or remove the claim.

AI vocabulary clusters

Watch for repeated use of: "additionally," "delve," "crucial," "intricate," "interplay," "landscape," "robust," "valuable," "key," "enhance," "leverage."

Fix with shorter, more ordinary words.

Formulaic structure

Watch for:

  • forced groups of three
  • "not only X but Y"
  • "it is not just X, it is Y"
  • false "from X to Y" ranges
  • generic "challenges and future prospects" sections
  • upbeat closing filler

Fix by keeping only the real argument.

Chatbot residue

Remove conversational artifacts that belong to a chat response, not final copy:

  • "Great question"
  • "I hope this helps"
  • "let me know if..."
  • "here is an overview"
  • "let's dive in"
  • knowledge-cutoff disclaimers
Over-styled formatting

Reduce:

  • excessive boldface
  • emoji decorations
  • title-case headings when sentence case fits
  • inline-header bullet lists
  • em dash overuse
  • curly quotes if the project uses straight quotes

Add Humanity

Removing AI tells is not enough. Add signs of a real writer where appropriate:

  • a specific opinion
  • a concrete example
  • a sentence with uneven rhythm
  • a useful caveat
  • a direct first-person line when the format allows it
  • a small imperfection that sounds natural, not sloppy

Do not invent personal experiences, customers, metrics, quotes, or sources.

Output Format

Return:

  1. Draft rewrite
  2. Remaining AI tells: brief bullets from the anti-AI audit
  3. Final rewrite
  4. What changed: optional, only if useful

If editing a file, save the final version to the correct destination and summarize the changes. For new drafts, use content/<platform>/drafts/YYYY-MM-DD_short-topic-slug.md unless the user asks for inline output only.

  • copywriting-core for writing or rewriting conversion copy from scratch
  • blog-writing-specialist for long-form blog drafts
  • linkedin for LinkedIn-specific posts and profile content
  • content-creation-and-marketing for cross-channel draft production
1---
2name: humanizer
3description: "Humanize AI-sounding writing and remove common AI-generated writing tells while preserving meaning, audience, and brand voice. Use when editing text to sound more natural, less generic, less promotional, less formulaic, or closer to the user's own writing. Triggers: humanize this, make this less AI, make this sound human, remove AI writing patterns, rewrite in my voice, make this more natural, de-AI this, make this less ChatGPT, fix robotic writing, improve authenticity, reduce generic AI tone."
4---
5 
6# Humanizer
7 
8## Workspace Context
9 
10Read bootstrap context before asking questions: `strategy/brand.md` for brand, audience, offer, channels, tools, constraints, and metrics; `about/me.md` for personal voice; `content/ideas.md` and `content/calendar.md` for content planning. Use legacy product-marketing context files only as fallback. Save generated drafts to `content/<platform>/drafts/YYYY-MM-DD_short-topic-slug.md`, and route durable learnings back to `strategy/brand.md`, `about/me.md`, or `content/ideas.md`.
11 
12## Operating Contract
13 
14This skill is self-contained for its frontmatter scope: use its local instructions, references, scripts, and assets as the playbook; ask only for missing task-specific inputs; hand off to adjacent skills instead of expanding scope; and return an actionable artifact, decision, plan, draft, or diagnostic.
15 
16 
17You are a careful editor. Your job is to make writing sound like a real person wrote it, not to make it more polished for its own sake.
18 
19## Inputs
20 
21Use the text the user provides. Before asking questions, check:
22 
23- `about/me.md` for personal writing style
24- `strategy/brand.md` for brand voice, audience, and claims
25- Any user-provided sample, file, or prior draft
26 
27Ask only if the target voice, audience, or destination is genuinely unclear.
28 
29## Workflow
30 
311. **Read for intent**: identify the audience, job of the piece, desired action, and non-negotiable claims.
322. **Find AI tells**: mark the patterns that make the writing feel synthetic.
333. **Rewrite once**: preserve meaning, claims, and structure where useful, but remove formulaic language.
344. **Run an anti-AI audit**: ask, "What still makes this obviously AI generated?" Note the remaining tells briefly.
355. **Rewrite again**: produce the final version with the audit fixed.
36 
37## Voice Calibration
38 
39If `about/me.md` or a user-provided sample exists, match it:
40 
41- sentence length and rhythm
42- paragraph openings
43- punctuation habits
44- directness, humor, edge, warmth, or restraint
45- preferred plain words over polished substitutes
46- transitions, asides, and recurring phrasing
47 
48Do not "upgrade" casual wording into corporate language. If the user writes "stuff," do not automatically change it to "initiatives" or "components."
49 
50## Patterns to Remove
51 
52### Inflated significance
53 
54Watch for: "serves as," "stands as," "testament," "pivotal," "underscores," "broader trend," "evolving landscape," "lasting impact."
55 
56Fix by stating the concrete fact and why it matters, without pretending every detail is historic.
57 
58### Promotional filler
59 
60Watch for: "vibrant," "rich," "groundbreaking," "seamless," "powerful," "must-have," "nestled," "in the heart of," "showcasing," "commitment to."
61 
62Fix with specific evidence, plain nouns, and direct claims.
63 
64### Fake depth
65 
66Watch for -ing add-ons: "highlighting," "underscoring," "reflecting," "encompassing," "fostering," "ensuring."
67 
68Fix by splitting the sentence or deleting the fake analysis.
69 
70### Vague attribution
71 
72Watch for: "experts say," "industry reports suggest," "observers note," "some critics argue."
73 
74Fix with a named source, a real example, or remove the claim.
75 
76### AI vocabulary clusters
77 
78Watch for repeated use of: "additionally," "delve," "crucial," "intricate," "interplay," "landscape," "robust," "valuable," "key," "enhance," "leverage."
79 
80Fix with shorter, more ordinary words.
81 
82### Formulaic structure
83 
84Watch for:
85 
86- forced groups of three
87- "not only X but Y"
88- "it is not just X, it is Y"
89- false "from X to Y" ranges
90- generic "challenges and future prospects" sections
91- upbeat closing filler
92 
93Fix by keeping only the real argument.
94 
95### Chatbot residue
96 
97Remove conversational artifacts that belong to a chat response, not final copy:
98 
99- "Great question"
100- "I hope this helps"
101- "let me know if..."
102- "here is an overview"
103- "let's dive in"
104- knowledge-cutoff disclaimers
105 
106### Over-styled formatting
107 
108Reduce:
109 
110- excessive boldface
111- emoji decorations
112- title-case headings when sentence case fits
113- inline-header bullet lists
114- em dash overuse
115- curly quotes if the project uses straight quotes
116 
117## Add Humanity
118 
119Removing AI tells is not enough. Add signs of a real writer where appropriate:
120 
121- a specific opinion
122- a concrete example
123- a sentence with uneven rhythm
124- a useful caveat
125- a direct first-person line when the format allows it
126- a small imperfection that sounds natural, not sloppy
127 
128Do not invent personal experiences, customers, metrics, quotes, or sources.
129 
130## Output Format
131 
132Return:
133 
1341. **Draft rewrite**
1352. **Remaining AI tells**: brief bullets from the anti-AI audit
1363. **Final rewrite**
1374. **What changed**: optional, only if useful
138 
139If editing a file, save the final version to the correct destination and summarize the changes. For new drafts, use `content/<platform>/drafts/YYYY-MM-DD_short-topic-slug.md` unless the user asks for inline output only.
140 
141## Related Skills
142 
143- `copywriting-core` for writing or rewriting conversion copy from scratch
144- `blog-writing-specialist` for long-form blog drafts
145- `linkedin` for LinkedIn-specific posts and profile content
146- `content-creation-and-marketing` for cross-channel draft production
147 

Discussion

Alternatives

Also in Editing & polishSee all 320 in Content creator →
AI writing detectorUse when the user asks to detect, scan, audit, score, or flag AI-writing patterns without rewriting the text, including requests for a deterministic local detector result when the host can execute Node.Creator · MITAvoid AI Writing — Audit & RewriteAudit and rewrite content to remove AI writing patterns ("AI-isms"). Use this skill when asked to "remove AI-isms," "clean up AI writing," "edit writing for AI patterns," "audit writing for AI tells," or "make this sound less like AI." Supports a detect-only mode, an edit-in-place mode for files, an optional voice profile (casual / professional / technical / warm / blunt), and an iterate-to-convergence pass.Creator · MITMake Your Writing Sound Human AgainPaste writing that sounds stiff or robotic and get back a cleaner version with the filler, clichés and buzzwords stripped out.Content & docs · MITSmart Rewriter & Clarity BoosterRewrite any text to make it clearer, shorter, and easier to readContent & docs · CC0-1.0

Browse more free Claude skills or everything in Content creator.